website-assistant/咨询访客

7.6 Chat Visitors

The Chat Visitors page displays all visitors who have triggered AI Customer Service.
Unlike the “Page Visitors” section, this module focuses specifically on:

  • Who opened customer service?
  • Who started a conversation?
  • Did they use the Knowledge Base (KB)?
  • What questions did they ask?
  • What path did they follow before and after chatting?

This is one of the most important modules for conversion analysis.

1. Page Overview

After entering Website Assistant → Chat Visitors, you will see a list of all visitors who have triggered customer service.

The list contains:

Field Description
Visitor ID (CID) Anonymous unique visitor identifier
Session Count Number of times the visitor triggered CS
First Chat Time Timestamp of the visitor’s first CS interaction
Recent Chat Time Timestamp of the latest conversation
Message Count Total messages in the current conversation
KB Used Whether a knowledge-base search occurred
Source Page Page where customer service was triggered
Device Mobile / PC / Tablet
Location Geographic location of the visitor

This helps you understand:

  • Where users come from
  • What they want to know
  • Whether they continued browsing after chat
  • Whether the Knowledge Base effectively answered questions

2. Chat Details

Clicking any visitor opens the Chat Details page, which includes:

1. Chat Messages

Displays the complete conversation:

  • Visitor questions
  • Knowledge Base hits (with citation excerpts)
  • AI responses
  • Whether fallback was triggered
  • Multi-turn dialog context
  • Uploaded images (if any)

Chat messages support:

  • Time sorting
  • Scroll loading
  • Quick search
  • Filter by KB-hit messages

2. Conversation Metadata

Includes:

  • Chat start time
  • Chat end time
  • Chat duration
  • Model used (GPT / DeepSeek / Local model)
  • Whether FAQ was triggered
  • Whether KB articles were matched
  • Matched KB document fragments with similarity scores

This is very important for evaluating Knowledge Base quality.

3. Trigger Path

Shows the visitor's behavior path before and after interacting with customer service:

Home → Product Page → Customer Service → Question Asked → Documentation → Exit

This helps you identify:

  • Where visitors encounter confusion
  • Whether their questions were resolved
  • Whether they continued browsing or bounced immediately

3. Filtering Visitors

You can filter chat visitors by:

  • Date range (Today / Yesterday / This Week / Custom)
  • Device type
  • Geographic region
  • Whether Knowledge Base was used
  • Source page
  • Multi-turn conversation vs. single-sentence exit
  • Model type (GPT, DeepSeek, Local)
  • Chat keywords (e.g., "price", "installation error")

Examples:

  • Visitors who asked only one question and left
  • High-value users who frequently trigger customer service
  • Visitors who rely only on FAQ
  • Whether the Knowledge Base answers most questions

4. Knowledge Base & Customer Service Insights

When a visitor asks a question, Pop attempts:

  1. Use Knowledge Base for the first response
  2. If no KB hit, fallback to:
    • Fixed fallback text, or
    • AI-generated response

The system records:

  • Whether KB answered the question
  • Hit article title
  • Hit paragraph excerpt
  • Similarity scores
  • Whether the visitor continued questioning
  • Whether the issue was resolved

You can use this information to improve your KB:

Situation Recommended Action
Many visitors ask the same question Add related documents to the KB
KB hit but users continue asking KB content needs refinement
Many fallback responses Improve FAQ or expand KB
Many chats but no conversion Optimize welcome message or trigger position

5. Chat Quality Analysis

Pop provides intelligent indicators, such as:

1. Chat Success Rate

Whether users continue browsing after receiving an answer.

2. Keyword Heatmap

Frequently asked topics like:

  • “pricing”
  • “how to install the client”
  • “error 401”

Useful for improving KB and optimizing content.

3. Model Performance

Includes:

  • Average response time
  • KB hit rate
  • FAQ hit rate
  • Fallback count
  • Visitor satisfaction (future rating support)

6. Summary

The Chat Visitors module helps you deeply understand:

  • Why users come to consult
  • What they ask
  • Whether the Knowledge Base is effective
  • Whether AI Customer Service improves retention
  • Which users deserve follow-up

It is a key module for optimizing conversion rate, user experience, and Knowledge Base quality.

Previous page
Website Visitors
Next Page
AI Customer Service